The Effectiveness of Financial Incentives for Health

The Effectiveness of Financial Incentives for HealthBehaviour Change: Systematic Review and Meta-AnalysisEmma L. Giles1*, Shannon Robalino1, Elaine McColl2, Falko F. Sniehotta1, Jean Adams1

1 Institute of Health and Society, Newcastle University, Newcastle upon Tyne, Tyne and Wear, United Kingdom, 2 Newcastle Clinical Trials Unit, Institute of Health and

Society, Newcastle University, The Medical School, Framlington Place, Newcastle upon Tyne, Tyne and Wear, United Kingdom

Abstract

Background: Financial incentive interventions have been suggested as one method of promoting healthy behaviourchange.

Objectives: To conduct a systematic review of the effectiveness of financial incentive interventions for encouraging healthybehaviour change; to explore whether effects vary according to the type of behaviour incentivised, post-interventionfollow-up time, or incentive value.

Data Sources: Searches were of relevant electronic databases, research registers, www.google.com, and the reference listsof previous reviews; and requests for information sent to relevant mailing lists.

Eligibility Criteria: Controlled evaluations of the effectiveness of financial incentive interventions, compared to nointervention or usual care, to encourage healthy behaviour change, in non-clinical adult populations, living in high-incomecountries, were included.

Study Appraisal and Synthesis: The Cochrane Risk of Bias tool was used to assess all included studies. Meta-analysis wasused to explore the effect of financial incentive interventions within groups of similar behaviours and overall. Meta-regression was used to determine if effect varied according to post-intervention follow up time, or incentive value.

Results: Seventeen papers reporting on 16 studies on smoking cessation (n = 10), attendance for vaccination or screening(n = 5), and physical activity (n = 1) were included. In meta-analyses, the average effect of incentive interventions was greaterthan control for short-term (#six months) smoking cessation (relative risk (95% confidence intervals): 2.48 (1.77 to 3.46);long-term (.six months) smoking cessation (1.50 (1.05 to 2.14)); attendance for vaccination or screening (1.92 (1.46 to 2.53));and for all behaviours combined (1.62 (1.38 to 1.91)). There was not convincing evidence that effects were different betweendifferent groups of behaviours. Meta-regression found some, limited, evidence that effect sizes decreased as post-intervention follow-up period and incentive value increased. However, the latter effect may be confounded by the former.

Conclusions: The available evidence suggests that financial incentive interventions are more effective than usual care or nointervention for encouraging healthy behaviour change.

Trial Registration: PROSPERO CRD42012002393

Citation: Giles EL, Robalino S, McColl E, Sniehotta FF, Adams J (2014) The Effectiveness of Financial Incentives for Health Behaviour Change: Systematic Reviewand Meta-Analysis. PLoS ONE 9(3): e90347. doi:10.1371/journal.pone.0090347

Editor: Hamid Reza Baradaran, Iran University of Medical Sciences, Iran (Islamic Republic of)

Received December 5, 2013; Accepted January 28, 2014; Published March 11, 2014

Copyright: � 2014 Giles et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permitsunrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This work is produced under the terms of a Career Development Fellowship research training fellowship issued by the NIHR to JA. The views expressedare those of the authors and not necessarily those of the NHS, The National Institute for Health Research or the Department of Health. JA & ELG are funded in part,and FFS is funded in full by Fuse: the Centre for Translational Research in Public Health, a UKCRC Public Health Research Centre of Excellence. Funding for Fusefrom the British Heart Foundation, Cancer Research UK, Economic and Social Research Council, Medical Research Council, the National Institute for HealthResearch, under the auspices of the UK Clinical Research Collaboration, is gratefully acknowledged. The funders had no role in study design, data collection andanalysis, decision to publish, or preparation of the manuscript.

Competing Interests: Jean Adams is a PLOS ONE Editorial Board Member; this does not alter our adherence to PLOS ONE Editorial policies and criteria. No otherauthors have any conflicts of interest.

* E-mail: [email protected]

Introduction

Despite consistent efforts to encourage uptake of healthy

behaviours [1,2], unhealthy behaviours remain common in

developed countries [3]. Financial incentives have been suggested

as one method of promoting healthy behaviour change.

Individual decisions to engage in behavioural options are

influenced by beliefs about the likely consequences of performing

those behaviours [4]. Individuals commonly hold inconsistent

preferences for outcomes occurring at different points in the

future, and for outcomes that are more or less certain. In general,

outcomes that will occur in the near future or with more certainty,

are valued more than those in the distant future or with less

PLOS ONE | www.plosone.org 1 March 2014 | Volume 9 | Issue 3 | e90347

www.google.com

http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42012002393#.Uv7D-7RFA7A

http://creativecommons.org/licenses/by/4.0/

certainty [4]. Whilst anticipated health gains of healthy behaviours

are often delayed in time and are uncertain (e.g. reduced risk of

disease in the future), the financial and opportunity costs can be

immediate and certain (e.g. giving up leisure time to take part in

physical activity) [5]. As these immediate, certain costs are often

‘dis-valued’ more than the delayed, uncertain health benefits are

valued, individuals make a ‘rational’ choice to pursue unhealthy

behaviours. It is hypothesised that health promoting financial

incentive interventions (HPFI) provide near-immediate and

certain rewards for, or reduce the immediate costs of, health-

promoting behaviours, and so change the reward structure

associated with these behaviours making them more attractive to

individuals [5].

The complexities of HPFI and the challenges of defining them

have been previously acknowledged [6,7]. However, incentive

interventions share in common that they offer motivating rewards

contingent on behavioural performance [6,8]. Here we define

HPFI as cash or cash-like rewards (e.g. vouchers that can be

exchanged for goods or services) or penalties (e.g. reductions in

welfare benefits), provided contingent on performance of healthy

behaviours.

It is commonly suggested that HPFI are more useful for

encouraging simple one-off behaviours, such as attendance for

vaccinations, than more complex sustained behaviour change,

such as smoking cessation [9–11]. However, we are not aware of

any systematic evidence synthesis that has arrived at this

conclusion, but it may be related to a common concern that the

effects of HPFI diminish quickly after incentives are withdrawn

[12,13], meaning that any behaviour change achieved is unlikely

to be sustained. A previous review did conclude that external

rewards can reduce an individual’s internal motivation to pursue

behaviour change, such that they become dependent on the

reward, rather than any personal desire to pursue the healthy

behaviour [14]. However, this finding is based on laboratory-

based research and may not be generalisable to community

settings.

Some authors have suggested that HPFI may be more suitable

for, or attractive to, individuals living in more deprived

circumstances [15,16]. However, variations in effectiveness of

HPFI across population groups have not been systematically

explored. Overall, little is known about what makes an effective

HPFI, in terms of value, format or other characteristics of the

incentive, behaviour, or recipient.

A number of reviews, using both systematic and other methods,

have now been conducted on the effects of HPFI [11,17–23].

However, these focus on single, specific behaviours rather than

exploring the full range of healthy behaviours [17,21–23]; are

restricted to developing countries where absolute financial

hardship may be more common than in developed countries

[20]; or use non-systematic methods for searching and screening,

meaning that findings may be biased [11,18,19].

We aimed to fill the gaps identified by conducting a systematic

review of primary studies exploring the effectiveness of HPFI

compared to non-intervention or usual care, to encourage uptake

of any healthy behaviours, in non-clinical adult populations living

in high-income countries. We also explored whether the effects of

HPFI varied according to the type of behaviour incentivised,

follow-up time after incentive withdrawal, or the value or format of

the incentive itself.

Methods

The protocol (see File S1) for this review was published in full

[24] and registered with PROSPERO before searching com-

menced (Registration no. 2012:CRD42012002393). Although a

number of the original research questions could not be answered

due to limited data availability, there were no substantive

variations from the protocol. The review is reported according

to the Preferred Reporting Items for Systematic Reviews

(PRISMA) guidelines (see Checklist S1) [25].

Information sourcesRelevant electronic databases of peer-reviewed literature were

searched from the earliest date available to April 2012. These

were: Medline, Embase, Science Citation Index, Cumulated Index

to Nursing and Allied Health Literature (CINAHL), Social

Science Citation Index, PsycINFO, Applied Social Science Index

and Abstracts, International Bibliography for the Social Sciences

and The Cochrane Library (including DARE, CENTRAL, HTA,

and NHS EDD). The search strategy combined relevant terms for

‘incentives’, ‘behaviour’ and ‘behaviour change’. An example of

the full electronic search strategy used in MEDLINE is provided in

File S2 and this was adapted, as appropriate, for other databases.

Manual searches of online research registers (Current Con-

trolled Trials, clinicaltrials.gov) were conducted alongside searches

of www.google.com. Relevant National Academic Mailing List

groups (Jiscmail) were also sent requests for relevant information.

The reference lists of relevant previous reviews [11,17,18,21–

23,26–46] and all papers meeting the inclusion criteria [47–63]

were also reviewed. Citation searches of included papers were

conducted using Science Citation Index and Social Science

Citation Index.

Endnote 66 was used to manage search results.

Eligibility criteriaWe searched for published and unpublished controlled evalu-

ations of the effectiveness of HPFI, compared to no intervention or

usual care, to encourage uptake of healthy behaviours in non-

clinical adult populations, living in high-income countries. The

inclusion criteria are described in full in Table S1. The inclusion of

all controlled study designs was as suggested by the Cochrane

Effective Practice and Organisation of Care Group (http://epoc.

cochrane.org/epoc-resource). In order to focus on the effect of

financial incentives on health behaviour change, only studies that

had behavioural outcomes were included. Studies that used

process markers of change only (e.g. weight loss, but not physical

activity or diet) were excluded. We restricted the review to studies

measuring behaviour change using objective, or validated self-

reported, methods to ensure high levels of validity. By ‘validated

self-report’ we mean non-objective measures that have previously

been reported to be valid compared to an objective measure. We

restricted the review to non-clinical populations to ensure

applicability to behaviour change in free-living ‘healthy’ adults

and so ensure maximum public health applicability. Many

‘incentive’ schemes offer participants a non-guaranteed reward

for behaviour change – e.g. entry into a lottery. Individuals differ

in their conceptualisations of risk and uncertainty, and so we

restricted the review to HPFI interventions provided with 100%

certainty to ensure that such differences did not confound the

results.

Study selection and data collectionAfter exclusion of duplicates, the title and abstract of all

retrieved papers were screened by one researcher (ELG) to exclude

obviously irrelevant papers. The full texts of remaining papers

were independently screened by two researchers (ELG & JA) to

identify those meeting the inclusion criteria. Any disagreements

were resolved by discussions.

Systematic Review of Financial Incentives


www.google.com

http://epoc.cochrane.org/epoc-resource

http://epoc.cochrane.org/epoc-resource

Data was extracted independently by two researchers (ELG &

JA) using a pro-forma developed for this purpose. Extracted

information included: bibliographic details, information on

participants, HPFI interventions, comparators, outcomes, study

design, and results. Incentive interventions were described using a

framework for this purpose [6]. Disagreements were resolved by

discussion.

The inclusion criteria restricted HPFI to those that would

definitely be provided if behaviour change occurred. The value of

these certain HPFI, over and above participant payments, were

identified and converted into 2011 US$ to allow comparisons

between studies (http://www.measuringworth.com/ppowerus/).

Uncertain, chance incentives (e.g. entry into lotteries), were offered

alongside certain incentives in some studies. As the probability of

winning, and the value of winnings, was often not clear, these were

not included in calculations of total incentive value.

Risk of biasRisk of bias in included studies was assessed independently by

two researchers (ELG & JA) using the Cochrane Risk of Bias

Review Guidelines [64]. Disagreements were resolved by discus-

sion.

Synthesis of resultsStudies examining the effects of HPFI for similar behaviours

were grouped together in a tabular summary for narrative

synthesis.

Incentives were described, using a framework for this purpose

[6], in terms of: direction (reward or penalty), form (cash, vouchers

or goods), magnitude, whether incentives were certain only or also

included chance components, target behaviour, frequency of

reward (all or some instances incentivised), immediacy of reward

in relation to behaviour, schedule (fixed or variable), and

recipient(s) of incentives (individuals or groups). Where studies

reported more than one relevant comparison (e. g. multi-arm trials

comparing a number of different incentive values to control), all

were identified and described.

For all groups of behaviours where more than one relevant

comparison was present and sufficient data were available, meta-

analysis was undertaken by group. Where more than one

intervention arm from a single study was included in a single

meta-analysis, the control group was divided in proportion to the

relative sizes in each intervention arm to avoid double counting

[65]. Many studies on smoking cessation included a number of

different follow-up points. Meta-analyses of smoking cessation

studies were performed for medium (#six months) and longer

term (.six months) follow-up points and included only the longest

follow-up point in each category from each study.

In addition to meta-analyses by behavioural group, an overall

meta-analysis including all comparisons from all included studies,

where data was available, was also performed. This was restricted

to only the longest follow-up point from studies including multiple

follow-ups.

Throughout, random-effects meta-analysis was conducted using

Cochrane Collaboration Review Manager 5.1. Risk ratios (RR)

and 95% confidence intervals (CI) were calculated for use in forest

plots. Where there was evidence of a high level of heterogeneity

(i.e. I2.75%) [66], further sub-group analyses by intervention

design were explored. Contour enhanced funnel plots were drawn

using Comprehensive Meta-Analysis 2.0 to assess potential

publication bias.

Meta-regression was conducted within the same groups as meta-

analyses, to explore whether log transformed study RR varied by

incentive value, or (where appropriate) follow-up period. No other

characteristics of interventions or participants were reported

consistently enough to allow exploration of the effects of these

on RR. Unrestricted maximum likelihood mixed-effects meta-

regression was conducted in Comprehensive Meta-Analysis 2.0

and meta-regression plots, with points proportional in size to

comparison weights drawn.

Results

The full text of three papers that were potentially relevant could

not be located and were excluded from the review [67–69]. The

full text of 350 papers were screened and a total of 17 papers met

the inclusion criteria and were included in the review (Figure 1).

Two papers reported data from different follow up points for the

same study, leaving 16 included studies [51,52].

The characteristics of included studies are summarised in Table

S2. Of the 16 studies, ten studies focused on smoking cessation

[48,50–54,59–63], five on attendance for vaccination or screening

[47,55–58], and one on physical activity [49]. All included studies

were randomised controlled trails (RCTs) or cluster RCTs. All 14

studies which provided information on location were conducted in

the USA [47–56,58,61–63]. Authors of the remaining studies were

based in the USA and it is likely that participants were too

[57,59,60].

Most HPFI offered were cash rewards [48,49,51,52,57,59–63]

and/or vouchers exchangeable for a specific range of goods or

services [47,55,56]. Two studies [53,54] used deposit contracts

where participants made cash deposits at the start of the study

which were only returned in the event of successful behaviour

change – resulting in potential financial penalties. Two studies also

included additional uncertain rewards contingent on behaviour

change (e.g. entry into lotteries) in addition to certain rewards

[50,58].

The total value of certain financial incentives that study

participants could receive for successful behaviour change, over

and above any payments for study participation, ranged from

$5.16 [57], to $786 (in 2011 $US) [62].

Intervention periods in the smoking cessation studies ranged

from two weeks [59,60], to 24 months [54], with post-intervention

follow-up periods ranging from four weeks [59,60] to 24 months

[50–52,54]. Most studies on attendance for vaccination or

screening involved a reward for one-off attendance with no

prolonged intervention or follow-up period [47,55–57]. One study

assessed repeated attendances for a series of injections over a 24

week period with incentives provided for each attendance [58].

The physical activity study had an intervention period of four

weeks with final follow-up immediately following the intervention

period [49].

The risk of bias in included studies was low or unclear in most

areas in most studies (Figure 2). Allocation sequence and allocation

concealment, together with possible selection bias (the main source

of ‘other’ bias arising from using volunteer samples), were the main

potential sources of bias. The risk of bias was high in relation to

allocation sequence, allocation concealment, and baseline charac-

teristics in one study [47]. This study was not included in meta-

analyses or meta-regression due to insufficient data being

presented on numbers of participants in each group, and details

of outcomes in each group. Attempts to contact the authors were

unsuccessful.

Overall, 15 of the 16 studies met the criteria for inclusion in

meta-analyses and meta-regressions: all ten of the studies on

smoking cessation, four out of five of the studies on attendance for

vaccination and screening, and the single study on physical activity

– although this latter study was only included in the analysis of all



http://www.measuringworth.com/ppowerus/

behaviours combined. The remaining study on attendance for

screening was excluded due to insufficient data [47].

Smoking cessationMeta-analyses of 13 comparisons from eight studies on smoking

cessation which reported outcomes #six months follow-up,

revealed an average RR (95% CI) of 2.48 (1.77 to 3.46)

(Figure 3) in favour of incentives. An I2 of 21%, indicating low

evidence of heterogeneity [66], was not explored further. Meta-

regression of this group of studies revealed no evidence that study

RR was associated with follow-up time (beta (95%CI): 20.003

(20.01 to 0.003); Figure 4) or total incentive value (beta (95%CI):

20.003 (20.001 to 0.0008); Figure 5).

Six studies, including eight comparisons, were included in meta-

analysis of the effect of financial incentives for smoking cessation

for follow-ups .six months. This revealed an average RR (95%

CI) of 1.50 (1.05 to 2.14) (Figure 6). An I2 of 76% indicated high

evidence of heterogeneity. Subgroup analyses suggested that the

average effect of cash-only financial incentives (RR (95%CI): 1.57

(1.06 to 2.32) was greater than that for other formats (RR

(95%CI): 1.16 (0.45 to 2.94) and that the latter was not statistically

significant. Although high heterogeneity remained in the cash-only

sub-group (I2 = 83%), all other sub-group analyses resulted in

inclusion of groups containing only one comparison.

Meta-regression showed no evidence that the log transformed

RR of financial incentives for smoking cessation with .six months

follow-up varied by follow-up period (coefficient (95%CI): 0.0005

(20.002 to 0.001); Figure 7). However there was some evidence

that log transformed RR increased as incentive value increased

(coefficient (95%CI): 0.001 (0.0002 to 0.003); Figure 8).

Contour enhanced funnel plots did not suggest any funnel plot

asymmetry for either group of smoking cessation comparisons

(Figure 9 and Figure 10) meaning that the risk of publication bias

was low.

Attendance for vaccination or screeningOf the five studies reporting on the use of financial incentives for

increasing attendance for vaccination and screening, one focused

on attendance at breast and cervical screening [47], two on

attendance for tuberculosis (TB) skin test reading [55,56] and one

each on attendance for influenza and hepatitis B vaccination

[57,58].

Figure 1. Flow diagram of study selection and exclusion.doi:10.1371/journal.pone.0090347.g001



Nine relevant comparisons from four studies were included in a

meta-analysis. The average RR (95%CI) was 1.92 (1.46 to 2.53)

(Figure 11) with evidence of a high level of heterogeneity

(I2 = 89%). Sub-group analyses suggested that cash plus other

motivational components (RR (95%CI): 2.75 (1.84 to 4.13)) may

be more effective than cash or vouchers alone (RR (95%CI): 1.77

Figure 2. Risk of bias in included studies.doi:10.1371/journal.pone.0090347.g002

Figure 3. Meta-analysis of financial incentives for smoking cessation (follow-up #six months).doi:10.1371/journal.pone.0090347.g003



(1.33 to 2.35)). Considerable heterogeneity remained in one subgroup

(I2 = 89%) but other approaches to subgroup analyses resulted in

subgroups containing only one comparison and were not pursued.

Meta-regression revealed no evidence that log transformed RR

varied by incentive value (coefficient (95%CI): 20.0004 (20.004

to 0.003); Figure 12). However, visual inspection of Figure 12

shows minimal variation in incentive values offered. A contour

enhanced funnel plot did not suggest any evidence of publication

bias (Figure 13).

Figure 4. Meta-regression of follow-up period on relative risk, smoking cessation (follow-up ,six months).doi:10.1371/journal.pone.0090347.g004

Figure 5. Meta-regression of incentive value on relative risk, smoking cessation (follow-up up ,six months).doi:10.1371/journal.pone.0090347.g005



Physical activityOnly one relevant comparison was included on physical activity

and meta-analysis was not undertaken for this behavioural group

[49]. This study used pedometers to measure average daily

physical activity over one-week periods and rewarded increases in

physical activity with increasing cash incentives. Over the four

Figure 6. Meta-analysis of financial incentives for smoking cessation (follow-up .six months).doi:10.1371/journal.pone.0090347.g006

Figure 7. Meta-regression of follow-up period on relative risk, smoking cessation (follow-up .six months).doi:10.1371/journal.pone.0090347.g007



Figure 8. Meta-regression of incentive value on relative risk, smoking cessation (follow-up of .six months).doi:10.1371/journal.pone.0090347.g008

Figure 9. Contour enhanced funnel plot, smoking cessation (follow-up #six months).doi:10.1371/journal.pone.0090347.g009



Figure 10. Contour enhanced funnel plot, smoking cessation (follow-up .six months).doi:10.1371/journal.pone.0090347.g010

Figure 11. Meta-analysis of financial incentives for attendance at vaccination and screening.doi:10.1371/journal.pone.0090347.g011



week intervention period, participants in the financial incentive

arm took part in an average of 16 more minutes of physical activity

per day than those in the control arm. This difference was

statistically significant.

All behavioursA total of 25 relevant comparisons were included in a meta-

analysis of all behaviours that included only the longest follow up

point from studies with multiple follow-ups. The average RR

Figure 12. Meta-regression of incentive value on relative risk, attendance at vaccination & screening.doi:10.1371/journal.pone.0090347.g012

Figure 13. Contour enhanced funnel plot, attendance at vaccination and screening.doi:10.1371/journal.pone.0090347.g013



(95%CI) was 1.62 (1.38 to 1.91) (Figure 14). Although an I2 of

84% suggested considerable heterogeneity, this was not explored

further.

Meta-regression showed some evidence that log transformed

RR decreased as post-intervention follow-up period increased

(coefficient (95%CI): 20.001 (20.002 to 20.0002); Figure 15) and

incentive value increased (coefficient (95%CI): 20.001 (20.002 to

20.0001); Figure 16). The funnel plot did not suggest clear

evidence of publication bias (Figure 17).

Discussion

Statement of principal findingsThis is the first systematic review which brings together

evidence on the effectiveness of financial incentive interventions

for encouraging uptake of the full range of health promoting

behaviours in non-clinical adult populations living in high-income

countries. A total of 17 papers reporting on 16 studies met the

inclusion criteria and were included in the review. These explored

the effect of HPFI on smoking cessation, attendance for

vaccinations or screening, and physical activity. Overall, this

Figure 14. Meta-analysis of financial incentives for all behaviours (latest follow-up point).doi:10.1371/journal.pone.0090347.g014



review found evidence that HPFI were more effective than no

intervention, or usual care, in changing behaviours. This was seen

for groups of similar behaviours (i.e. smoking cessation, attendance

for vaccinations or screening) as well as when all behaviours were

combined. There was no clear evidence that HPFI were more

effective for ‘simple’ behaviours (e.g. attendance for vaccination or

screening) than ‘complex’ ones (e.g. smoking cessation). Financial

incentive interventions took a range of formats and it was difficult

to draw conclusions on the most effective of these, particularly

given the lack of detailed information on the exact nature of

interventions and study participants, as well as the absence of

trials which have sought to determine if effects of interventions

vary according to socio-demographic characteristics. When all

behaviours were grouped together, there was some evidence that

effect decreased as post-intervention period increased and as total

incentive value increased. However, it is possible that the latter

effect was confounded by the former.

Strength and weaknesses of studies in this reviewThis review found few controlled studies exploring the effect of

HPFI. The studies that were found were restricted to a small

number of behaviours. Further, primary, controlled studies

exploring the effect of financial incentives on change in a range

of other health-related behaviours are required.

The studies included did not appear to be at high risk of bias,

but there were some areas that were consistently at greater risk of

Figure 15. Meta-regression of follow-up point on relative risk, all behaviours (latest follow-up point).doi:10.1371/journal.pone.0090347.g015

Figure 16. Meta-regression of incentive value on relative risk, all behaviours (latest follow-up point).doi:10.1371/journal.pone.0090347.g016



bias, particularly: allocation sequence generation and allocation

concealment. In most cases, information was not provided, rather

than it being clear that methods were weak. Future researchers

should consistently report trials according to existing reporting

guidance [70].

It appears that all 16 studies included in the review were US-

based, potentially limiting generalisability to other cultures and

contexts. Further studies based in other countries are required to

confirm that the effects reported here are generalisable to other

contexts.

The meta-regression plots revealed some gaps in the range of

incentive values and follow-up periods that have been explored. In

particular, few studies explored medium-size incentives (e.g. $40–

250) for encouraging attendance for vaccination and screening;

and few studies had post-intervention follow up periods beyond six

months.

Although we originally intended to explore if the effect of HPFI

varied according to recipient characteristics (e.g. age, gender,

socio-economic position), data was not reported in such a way to

allow this. Many other public health interventions are differentially

effective according to socio-demographic characteristics of partic-

ipants [71]. Further research is required to determine if HPFI are

particularly effective in some population groups.

Strengths and weaknesses of this reviewThis is the first systematic review and meta-analysis, that we are

aware of, that has explored the effect of financial incentives across

the full range of healthy behaviours in non-clinical settings in high-

income countries. Previous reviews have either focused on single

health behaviours [17,21–23], failed to use standard systematic

review methods [11,18,34], or have been limited to low and

middle-income countries [20].

Restriction to controlled study designs is recommended to

minimise risk of bias in conclusions and so increase confidence in

results (http://epoc.cochrane.org/epoc-resources). However, it

has been argued that other study designs can also contribute

useful information to reviews and an alternative approach to

evidence synthesis, for example a realist synthesis exploring the

context, mechanisms and outcomes of effective components of

financial incentive interventions for health behaviour change [72],

may help shed additional light on the financial incentive field. The

strict inclusion criteria, such as only including studies with

objective, or validated self-report behavioural outcome measures

and our focus on behaviour change (e.g. physical activity) rather

than proxies of this (e.g. body weight), further adds to the

confidence in the results; but similarly limits the number of studies

meeting the inclusion criteria.

Only studies comparing HPFI to usual care or no intervention

were included. Thus, the results indicate the effect of financial

incentives compared to minimal intervention. It is not, therefore,

clear how HPFI compare to other interventions. Given the

controversy associated with HPFI [73–76], society may prefer to

avoid the widespread use of HPFI if similarly effective alternative

interventions are available.

We used an extensive search strategy, including database

searches, expert recommendations of studies, and reference and

citation searches. As such, we are confident that we are unlikely to

have missed any relevant studies. However, it is difficult to

conclusively confirm this. In particular, three studies were not

fully screened for inclusion as the full papers could not be

retrieved [67–69].

We found considerable heterogeneity within some meta-

analyses. This likely reflects differences in methods, populations,

and interventions and is a reality of the type of intervention we

were studying [77]. We clearly presented the heterogeneity found,

whilst trying to choose appropriate sub-groups to limit it. We also

ensured that inclusion criteria were robust and checked that the

data was correct where in doubt (by contacting authors) before

undertaking meta-analyses [78].

Figure 17. Contour enhanced funnel plot, all behaviours (latest follow-up point).doi:10.1371/journal.pone.0090347.g017



http://epoc.cochrane.org/epoc-resources

Interpretation of findings and comparison to previousfindings

Unlike previous non-systematic reviews [11,18,34], we used

well-recognised systematic review methodology with clear inclu-

sion criteria, substantially reducing the risk of bias in our findings.

Unlike previous systematic reviews [17,21–23], our inclusion

criteria covered the full range of healthy behaviours in non-clinical

adult populations living in high-income countries. Together, these

represent significant improvements on previous reviews.

Similar to previous findings, we found evidence that financial

incentives are effective at encouraging health-promoting behav-

iours [21,23,34]. In the meta-regression including all behaviours,

there was some evidence that effectiveness may decrease over time

post intervention period. This has previously been reported

[17,21,23]. However, statistically significant effects persisted at

least until six months post-intervention follow-up in smoking

cessation studies, suggesting that effects do not suddenly drop off

once incentives are withdrawn. Many health promotion interven-

tions are associated with behaviour change in the short, but not

longer, term [79] and this problem is not unique to HPFI.

Previous authors have suggested that HPFI may be more

effective in changing one-off health behaviours (such as attendance

for vaccination and screening) than more complex behaviours

(such as smoking) [9,11,12,23]. This study did not find convincing

evidence of this with the average RR of incentives for attendance

at vaccinations and screening (RR 1.92 (1.46 to 2.53) being less

than that for medium term smoking cessation (RR 2.48 (1.77 to

3.46), but greater than that for longer term smoking cessation (RR

1.50 (1.05 to 2.14). It is possible that the distinction between ‘one-

off’ and ‘complex’ behaviours is a false dichotomy in the context of

HPFI – with incentives for smoking cessation rewarding a series of

one-off behaviours. However, McEachan et al (2010) provide

evidence that smoking and attending for screening are considered

conceptually different on a number of dimensions by both ‘experts’

and members of the public [80]. Further work could usefully

explore whether the effectiveness of HPFI varies according to the

behavioural dimensions identified by McEachan et al (2010) – but

not enough data was available for such analysis in the current

review.

Meta-regression of all behaviours combined showed some

evidence that effects decreased as incentive value increased. As

incentive value was positively correlated with longest post-

intervention follow-up point (r = 0.44, p = 0.03), the finding in

relation to incentive value may be confounded by that related to

follow-up period. Unfortunately we were not able to conduct

multi-variate meta-regression to take account of this. Furthermore,

we found a positive relationship between inventive value and effect

size in smoking cessation studies with follow up of .six months,

suggesting that this effect is not consistent. Very weak evidence in

favour of larger value HPFI being more effective has been

previously reported [21]. However, other authors have suggested

that larger incentive values may be interpreted by recipients as

reflecting that the behaviour incentivised is somehow ‘risky’ and

thus that a payment is needed to offset this [81]. There are likely to

be complex relationships between incentive value and character-

istics of both incentivised behaviours and recipients that require

more detailed exploration.

Similar to previous reviews, we also found that the majority of

studies were US-based [34], and that insufficient evidence is

currently available to determine optimal incentive value, or format

for changing health behaviours [18]. The average RR reported

here are larger than some previously reported. For example, in

their original review of incentives as well as competitions for

smoking cessation, Cahill & Perera (2008) reported a pooled odds

ratio (95%CI) at six months follow up of 1.44 (1.01 to 2.04) [17],

(pooled odds ratios are not reported in the updated version of this

review) [82]. The comparable figures for the results in this review

were 3.12 (1.95 to 4.97) (note this is an odds ratio for

comparability, but the figures shown elsewhere are risk ratios).

However, this previous review included both competitions, as well

as incentives, and these differences in inclusion criteria may

explain the differences in results found.

Implications for policy and practiceFinancial incentives may be a useful addition to the behavioural

change toolkit, particularly for encouraging smoking cessation and

attendance for vaccination and screening. We did not find

convincing evidence that HPFI work better for changing short-

term, one-off behaviours than longer term, more complex

behaviours and HPFI should be considered across the spectrum

of healthy behaviours. Although there has been some previous

concern that the effects of HPFI may be short-lived once

incentives are withdrawn, we did not find convincing evidence

of this. Nor did we find convincing evidence that larger incentive

values are associated with greater behaviour change. This suggests

that small incentives may be effective, although it is not clear from

our results if larger incentives produce larger effects. However,

these issues have not been systematically investigated and it is not

clear what the most effective value, or format, of HPFI is.

Conclusion

The available evidence from controlled studies suggests that

HPFI are more effective than usual care or no intervention at

encouraging healthy behaviour change amongst non-clinical adult

populations living in high income countries. There was not

convincing evidence that HPFI are more effective for ‘simple’

compared to ‘complex’ behaviours. There was some evidence that

effects decrease as post-intervention follow-up increases and as

incentive value increases. However, the available evidence is

substantially limited, particularly in relation to the range of

behaviours studied.

Supporting Information

File S1 Protocol.

(PDF)

File S2 Search strategy for identification of studies for financial

incentives.

(PDF)

Table S1 Inclusion criteria.

(PDF)

Table S2 Characteristics of included studies.

(PDF)

Checklist S1 PRISMA checklist.

(PDF)

Author Contributions

Conceived and designed the experiments: JA FFS EM. Performed the

experiments: ELG JA SR. Analyzed the data: JA ELG. Contributed

reagents/materials/analysis tools: ELG JA FFS SR EM. Wrote the paper:

ELG JA EM FFS SR.



References

1. Department of Health (1998) Our healthier nation: a contract for health.

London: Her Majesty’s Stationery Office.

2. Department of Health (2004) Choosing health: making healthier choices easier.London: Her Majesty’s Stationery Office.

3. Eastwood P (2009) Health survey for England 2008: trend tables.

4. Mazur JE (1987) An adjusting procedure for studying delayed reinforcement. In:

Commons ML, Mazur JE, Nevin JA, Rachlin H, editors. Quantitative analysisof behaviour The effect of delay and intervening events on reinforcement value.

Hillsdale, New Jersey: Erlbaum.

5. Lowenstein G, Brennan T, Volpp K (2007) Asymmetric paternalism to improvehealth behaviors. Journal of the American Medical Association 298: 2415–2417.

6. Adams J, Giles EL, McColl E, Sniehotta F (2013) Carrots, sticks, and health

behaviours: a framework for documenting the complexity of financial incentiveinterventions to change health behaviours. Health Psychology Review.

7. Johnston M, Sniehotta F (2010) Financial incentives to change patient

behaviour. Journal of Health Services Research and Policy 15: 131–132.

8. Abraham C, Michie S (2008) A taxonomy of behaviour change techniques usedin interventions. Health Psychology 27: 379–387.

9. Jochelson K (2007) Paying the patient: improving health using financial

incentives. London: King’s Fund.

10. Crockett R, Weinman J, Hankins M, Marteau T (2009) Time orientation andhealth-related behaviour: measurement in general population samples. Psychol-

ogy & Health 24: 333–350.

11. Sutherland K, Christianson JB, Leatherman S (2008) Impact of targetedfinancial incentives on personal health behavior: a review of the literature.

Medical Care Research and Review 65: 36S–78.

12. Oliver A, Marteau T, Ashcroft RE (2009) Can financial carrots improve health?Journal of Health Services Research & Policy 14: 1–2.

13. Petry NM (2010) Contingency management treatments: controversies and

challenges. Addiction 105: 1507–1509.

14. Deci E, Koestner R, Ryan R (1999) A meta-analytic review of experimentsexamining the effects of extrinsic rewards on intrinsic motivation. Psychological

Bulletin 125: 627–668.

15. Aveyard P, Bauld L (2011) Incentives for promoting smoking cessation: what westill do not know. Cochrane Database Syst Rev Apr13: ED000027 Avaliable:

http://www.thecochranelibrary.com/details/editorial/1052523/Incentives-for-promoting-smoking-cessation-what-we-still-do-not-know.html.

16. Cookson R (2008) Should disadvantaged people be paid to take care of their

health? Yes. BMJ 337.

17. Cahill K, Perera R (2008a) Competitions and incentives for smoking cessation.Cochrane Database of Systematic Reviews doi: 10.1002/14651858.CD004307.

pub3

18. Kane R, Johnson P, Town R, Butler M (2004) A structured review of the effectof economic incentives on consumers’ preventive behavior. American Journal of

Preventive Medicine 27: 327–352.

19. Kavanagh J, Stansfield C, Thomas J (2009) Incentives to improve smoking,physical activity, dietary and weight management behaviours: a scoping review

of the research evidence. London: Institute of Education, University of London.

20. Lagarde M, Haines A, Palmer N (2007) Conditional cash transfers for improvinguptake of health interventions in low- and middle-income countries: a systematic

review. JAMA 298: 1900–1910.

21. Paul-Ebhohimhen V, Avenell A (2008) Systematic review of the use of financialincentives in treatments for obesity and overweight. Obesity Reviews 9: 355–

367.

22. Prendergast M, Podus D, Finney J, Greenwell L, Roll J (2006) Contingencymanagement for treatment of substance use disorders: a meta-analysis.

Addiction 101: 1546–1560.

23. Wall J, Mhurchu C, Blakely T, Rodgers A, Wilton J (2006) Effectiveness of

monetary incentives in modifying dietary behavior: a review of randomized,controlled trials. Nutrition Reviews 64: 518–531.

24. Adams J, Giles EL, McColl E, Sniehotta F (2012) A systematic review of the use

of financial incentives to encourage uptake of healthy behaviours: protocol.Systematic Reviews 1: doi:10.1186/2046-4053-1181-1151.

25. Moher D, Liberati A, Tetzlaff J, Altman DG, The Prisma Group (2009)

Preferred reporting items for systematic reviews and meta-analyses: thePRISMA statement. PLoS Med 6: e1000097.

26. Achat H, McIntyre P, Burgess M (1999) Health care incentives in immunisation.

Aust N Z J Public Health 23: 285–288.

27. Avenell A, Broom J, Brown T, Poobalan A, Aucott L, et al. (2004) Systematicreview of the long-term effects and economic consequences of treatments for

obesity and implications for health improvement. Health Technol Assessment 8:1–182.

28. Bryant J, Bonevski B, Paul C, McElduff P, Attia J (2011) A systematic review and

meta-analysis of the effectiveness of behavioural smoking cessation interventionsin selected disadvantaged groups. Addiction 106: 1568–1585.

29. Cahill K, Perera R (2008) Quit and win contests for smoking cessation.

Cochrane Database of Systematic Reviews 3: CD004986.

30. Flodgren G, Eccles M, Shepperd S, Scott A, Parmelli E, et al. (2011) Anoverview of reviews evaluating the effectiveness of financial incentives in

changing healthcare professional behaviours and patient outcomes. CochraneDatabase of Systematic Reviews 6: CD009255.

31. Guiffrida A, Torgerson D (1997) Should we pay the patient? Review of financial

incentives to enhance patient compliance. BMJ 315: 703–707.

32. Hey K, Perera R (2005) Competitions and incentives for smoking cessation.

Cochrane Database of Systematic Reviews 2: CD004307.

33. Hey K, Perera R (2005) Quit and win contests for smoking cessation. Cochrane

Database of Systematic Reviews 2: CD004986.

34. Kavanagh J, Stansfield C, Thomas J (2009) Incentives to improve smoking,

physical activity, dietary and weight management behaviours: a scoping review

of the research evidence. London: Institute of Education.

35. Kavanagh J, Oakley A, Harden A, Trouton A, Powell C (2011) Are incentive

schemes effective in changing uoung people’s behaviour? A systematic review.

Health Education Journal 70: 192–205.

36. Leeks KD, Hopkins D, Soler R, Aten A, Chattopadhyay S, et al. (2010)

Worksite-based incentives and competitions to reduce tobacco use. A systematic

review. Am J Prev Med 38: S263–274.

37. Lussier JP, Heil SH, Mongeon JA, Badger GJ, Higgins ST (2006) A meta-

analysis of voucher-based reinforcement therapy for substance use disorders.

Addiction 101: 192–203.

38. Matson D, Lee J, Hopp J (1993) The impact of incentives and competitions on

participation and quit rates in worksite smoking cessation programs. Am J Health

Promot 7: 270–280.

39. Morgan O (2005) Approaches to increase physical activity: reviewing the

evidence for exercise-referral schemes. Public Health 119: 361–370.

40. Ogilvie DEM, Hamilton V, Petticrew M (2005) Promoting walking and cycling

as an alternative to using cars: a systematic review. BMJ 329: 1–5.

41. O’Malley GC, Baker PRA, Francis DA, Perry I, Foster C (2012) Incentive-based

interventions for increasing physical activity and fitness (protocol). Cochrane

Database of Systematic Reviews 1.

42. Peersman G, Harden A, Oliver S (1998) Effectiveness of health promotion

interventions in the workplace: a review. London: Health Education Authority.

43. Roux L, Pratt M, Tengs T, Yore M, Yanagawa T, et al. (2008) Cost-effectiveness

of community-based physical activity interventions. American Journal of

Preventive Medicine 35: 578–588.

44. Sabatino SA, Habarta N, Baron RC, Coates RJ, Rimer BK, et al. (2008)

Interventions to increase recommendation and delivery of screening for breast,

cervical, and colorectal cancers by healthcare providers - systematic reviews of

provider assessment and feedback and provider incentives. Am J Prev Med 35:

S67–S74.

45. Stone E, Morton S, Hulscher M, Maglione M, Roth E, et al. (2002)

Interventions that increase use of adult immunization and cancer screening

services: a meta-analysis. Annals of Internal Medicine 136: 641–651.

46. Yabroff KR, Mandelblatt JS (1999) Interventions targeted towards patients to

increase mammography use. Cancer Epidemiology, Biomarkers and Prevention

8: 749–757.

47. Debari RR, Servodidio C, Palomares M, Rodruguez-Furlow M, Staff I (2007)

Patient compliance/incentive study. Oncol Nurs Forum 34: 488.

48. Donatelle R, Prows S, Champeau D, Hudson D (2000) Randomised controlled

trial using social support and financial incentives for high risk pregnant smokers:

significant older supporter (SOS) program. Tobacco Control 9: 67–69.

49. Finkelstein E, Brown D, Brown D, Buchner DA (2008) A randomized study of

financial incentives to increase physical activity among sedentary older adults.

Prev Med 47: 182–187.

50. Glasgow R, Hollis J, Ary D, Boles S (1993) Results of a year-long incentives-

based worksite smoking-cessation program. Addict Behav 18: 455–464.

51. Jason LA, McMahon SD, Salina D, D.. Hedeker, Stockton M, et al. (1995)

Assessing a smoking cessation intervention involving groups, incentives and self-

help manuals. Behav Ther 26: 393–408.

52. Jason LA, Salina D, McMahon SD, Hedeker D, Stockton M (1997) A worksite

smoking intervention: a 2-year assessment of groups, incentives and self-help.

Health Educ Res 12: 129–138.

53. Jeffery R, Hellerstedt W, Schmid T (1990) Correspondence programs for

smoking cessation and weight control: a comparison of two strategies in the

Minnesota heart health program. Health Psychol 9: 585–598.

54. Jeffery RW, Forster JL, French SA, Kelder SH, Lando HA, et al. (1993) The

healthy worker project: a worksite intervention for weight control and smoking

cessation. Am J Public Health 83: 395–401.

55. Malotte C, Rhodes F, Mais K (1998) Tuberculosis screening and compliance

with return for skin test reading among active drug users. Am J Public Health 88:

792–796.

56. Malotte C, Hollingshead J, Rhodes F (1999) Monetary versus nonmonetary

incentives for TB skin test reading among drug users. Am J Prev Med 16: 182–

188.

57. Nowalk MP, Lin CJ, Toback Sl, Rousculp MD, Eby C, et al. (2010) Improving

influenza vaccination rates in the workplace. Am J Prev Med 38: 237–246.

58. Stitzer M, Polk T, Bowles S, T.. Kosten (2010) Drug user’s adherence to a 6-

month vaccination protocol: effects of motivational incentives. Drug Alcohol

Depend 107: 76–79.

59. Stitzer M, Bigelow G (1983) Contingent payment for carbon monoxide

reduction: effects of pay amount. Behav Ther 14: 647–656.



http://www.thecochranelibrary.com/details/editorial/1052523/Incentives-for-promoting-smoking-cessation-what-we-still-do-not-know.html

http://www.thecochranelibrary.com/details/editorial/1052523/Incentives-for-promoting-smoking-cessation-what-we-still-do-not-know.html

60. Stitzer M, Bigelow G (1985) Contingent reinforcement for reduced breath

carbon monoxide levels: target-specific effects on cigarette smoking. Addict

Behav 10: 345–349.

61. Volpp K, Levy AG, Asch DA, Berlin JA, Murphy JJ, et al. (2006) A randomized

controlled trial of financial incentives for smoking cessation. Cancer Epidemi-

ology, Biomarkers and Prevention 15: 12–18.

62. Volpp KG, Troxel AB, Pauly MV, Glick HA, Puig A, et al. (2009) A randomized

controlled trial of financial incentives for smoking cessation. The New England

Journal of Medicine 360: 699–709.

63. Windsor R, Lowe J, Bartlett E (1988) The effectivness of a worksite self-help

smoking cessation program: a randomized trial. J Behav Med 11: 407–421.

64. Thomas B, Ciliska D, Dobbins M, Micucci S (2004) A process for systematically

reviewing the literature: providing the research evidence for public health

nursing interventions. Worldviews on Evidence-Based Nursing 1: 17–84.

65. Higgins J, Green S (2011) Cochrane handbook for systematic reviews of

interventions version 5.1. [updated March 2011]. Available: www.cochrane-

handbook.org: The Cochrane Collaboration.

66. Higgins J, Thompson S, Deeks J, Altman D (2003) Measuring inconsistency in

meta-analyses. BMJ 327: 557–560.

67. National Business Group on Health (2007) Financial incentives for healthy

lifestyles: who’s doing it, what’s legal and where’s the evidence? Avaliable: www.

businessgrouphealth.org/toolkits/et_financialincentives.cfm. Accessed 2013

June.

68. Robinow A (2007) Consumer incentives: strong examples in action: patient

choice health care. Philadelphia, PA: National Business Coalition on Health/

Leapfrog Group/Bridges to Excellence Meeting, Advancing Value-Driven

Health Care: The Third Annual Incentives and Rewards Symposium.

69. Resnick JH (1974) Smoking behavior - motives and incentives. Behavior

Therapy 5: 711–712.

70. Schulz K, Altman D, Moher D (2010) CONSORT 2010 Statement: updated

guidelines for reporting parallel group randomised trials. BMJ 340: c332.71. White M, Adams J, Heywood P (2009) How and why do interventions that

increase health overall widen inequalities within populations? In: Babones S,

editor. Social inequality and public health. Bristol: Policy Press. pp. 65–82.72. Pawson R (2006) Evidence based policy: a realist perspective. London: Sage.

73. Gardner B, Davies A, McAteer J, Michie S (2010) Beliefs underlying UKparents’ views towards MMR promotion interventions: a qualitative study.

Psychol Health Med 15: 220–230.

74. Oliver A, Marteau T, Ashcroft R (2009) Can financial carrots improve health?J Health Serv Res Policy 14: 1–2.

75. Parke J, Ashcroft R, Brown R, Marteau T, Seale C (2013) Financial incentives toencourage healthy behaviours: an analysis of UK media coverage. Health

Expect 16: 292–304.76. Promberger M, Brown R, Ashcroft R, Marteau T (2011) Acceptability of

financial incentives to improve health outcomes in the UK and US samples.

J Med Ethics 37: 682–687.77. Cochrane Collaboration (2002) Module 13: diversity and heterogeneity.

78. Higgins JPT, Thompson SG (2008) Commentary: heterogeneity in meta-analysis should be expected and appropriately quantified. International Journal

of Epidemiology 37: 1158–1160.

79. Rothman AJ (2000) Toward a theory-based analysis of behavioral maintenance.Health Psychology 19: 64–69.

80. McEachan RRC, Lawton RJ, Conner M (2010) Classifying health-relatedbehaviours: exploring similarities and differences amongst behaviours. British

Journal of Health Psychology 15: 347–366.81. Gneezy U, Meier S, Rey-Biel P (2011) When and why incentives (don’t) work to

modify behavior. Journal of Economic Perspectives 25: 191–210.

82. Cahill K, Perera R (2011) Competitions and incentives for smoking cessation.Cochrane Database of Systematic Reviews 4: CD004307.



www.cochrane-handbook.org

www.cochrane-handbook.org

www.businessgrouphealth.org/toolkits/et_financialincentives.cfm

www.businessgrouphealth.org/toolkits/et_financialincentives.cfm

Documents

The Effectiveness of Financial Incentives for Health